The IT Industry’s 95% Problem

I just had a chance to read Chris O’Brien’s informative piece on VentureBeat titled “Evernote’s 5% problem offers a cautionary lesson to tech companies”. The article examines how Evernote ended up with a product that has too many features.

Based on an interview with former Evernote CEO Phil Libin, he was quoted as saying:

“What winds up happening at Evernote conferences is that people go and they say, ‘Oh, I love Evernote and I’ve been using it for years and now I realize I’ve only been using it for 5 percent of what it can do,’ ” Libin said. “And the problem is that it’s a different 5 percent for everyone. If everyone just found the same 5 percent, then we’d just cut the other 95 percent and save ourselves a lot of money. It’s a very broad usage base. And we need to be a lot better about tying it together. And I think we have. We’ve got a few things we’re launching over the next few months to help with that.”

This is the heart of what O’Brien is referring to as the “5% problem.”

Not to take issue with O’Brien’s great article, but I would suggest that this only adds up to a 5% problem if we are looking at this from Evernote’s perspective. If we look at this from the user’s perspective it strikes me that there’s a 95% problem. 95% of the features in Evernote create no value for any particular customer.

The only other product that seems to get away with the same ratio of superfluousness as software is fast food. That’s because most of the ingredients in burgers, fries and milkshakes are not what anyone would expect to find in a burger, a french fry or a milkshake.

Another thing…how did we get to such a low standard? 5%? Really? In fairness, I’ve heard this same contorted logic from many other vendors patting themselves on the back for just-as-bad 20% functionality usage. If a piece of software doesn’t have a majority of its capabilities being used by most of its users, I think that’s a problem right there. So, the benchmark should start at 50% and anything lower than that needs to trigger a career re-assignment for the product manager. 

I don’t mean to snipe at Libin or at Evernote. It’s just that O’Brien’s article highlights Evernote for what is really an industry-wide problem. After nearly three decades in the IT industry, the one truism I can rely on is that over-engineering software is a matter of when, not if. This not a cost-free problem. As O’Brien points out, expanding feature sets result in products that don’t have a core experience. In the case of Evernote, it’s become a personal productivity, group collaboration and enterprise information management tool. It’s a floor wax AND a dessert topping. The operative question is how many people want Evernote to be all these things at once?

Another problem that I find myself consistently highlighting to application/product managers is that perpetual functional expansion necessitates perpetuate UI remediation. There is a marked misunderstanding on the part of application/product managers that the blowback they’re getting on lousy UIs is most often the result of the unresolved prioritisation and presentation of ever-expanding feature sets. I find the people too often asking “how do I fix the UI” instead of asking “how did we break the UI.”

But the problem is persistent across the industry. Several years ago, Microsoft did an analysis of command use in Microsoft Word 2003. The top five commands – Paste, Save, Copy, Undo, Bold – accounted for 32% of total command usage! Furthermore they found:

“Beyond the top 10 commands or so the curve flattens out considerably. The percentage difference in usage between the No. 100 command (‘Accept Change’) and the No. 400 command (‘Reset Picture’) is about the same in differencebetween No. 1 and No. 11 (‘Change Font Size’).”

As software products initially emerge, it’s reasonable to assume that most of the functionality is used by most customers. Over time, the long tail of under-utilised features gets longer and flatter.

But is this just the nature of software? Is it something we just have to accept? I don’t believe so. But changing this requires application/product managers to shift their perspective of what constitutes user satisfaction.

Put simply, people are satisfied when the get the product they need, not the feature(s) they asked for. Sure, people want the features they asked for. But they don’t necessarily want the features that others asked for. The distinction is subtle, but clear. When we connect satisfaction to the product as a whole, superfluous capabilities are more clearly recognised as value destroying. 

So, how do we get over the “everyone uses a different x% of features” argument? We manage that problem by seeing this as an opportunity to compose functionality into products that resonate with the way users see themselves. Let me give you an example. Mariner Software makes, amongst other things, word processors. Notice the use of the plural. They have a word processor specific for the needs of screenwriters (Montage). They have a word processor for novelists (StoryMill). They have a word processor for journalist/bloggers (MacJournal). And they have a bare bones word processor that gets the main typed stuff done that most people at home, or school, contend with (Write).

Now Mariner Software would rightly take me to task for the use of the term “word processor”. But I’m using the term to highlight a broader point. I’m guessing that these products share significant functional overlap. But they vary based on unique needs of specifically defined audiences. That’s astute. They’re not telling the screenwriter to find their screenwriting features amongst the features that journalists need. They’re saying “here’s a tool just for you.” The strength of the idea is that their customers already self-identify as a screenwriter, a novelist, a journalist.

Of course, to pull this off application/product managers need to spend much more time identifying and understanding the nuanced behaviour of their customers than relying on the pace that they’re dropping new features to all their customers at once. But the payoff, in my opinion, is that we finally have a way of tackling the IT industry’s endemic 95% problem.